<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title>Self-Attention - 标签 - Luhong's Blog</title><link>https://blog.luhg.cn/tags/self-attention/</link><description>Self-Attention - 标签 - Luhong's Blog</description><generator>Hugo -- gohugo.io</generator><language>zh-CN</language><lastBuildDate>Tue, 09 Jun 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://blog.luhg.cn/tags/self-attention/" rel="self" type="application/rss+xml"/><item><title>03什么是Self-Attention</title><link>https://blog.luhg.cn/03%E4%BB%80%E4%B9%88%E6%98%AFself-attention/</link><pubDate>Tue, 09 Jun 2026 00:00:00 +0000</pubDate><author>Luhong</author><guid>https://blog.luhg.cn/03%E4%BB%80%E4%B9%88%E6%98%AFself-attention/</guid><description><![CDATA[<h1 id="self-attentiontransformer-里最聪明的那个大脑区域">Self-Attention：Transformer 里最聪明的那个“大脑区域”</h1>
<blockquote>
<p>如果你只想搞懂 Transformer 里的一个东西，那就搞懂 Self-Attention。
其他组件都可以看作是它的“助理”。</p>
</blockquote>
<hr>
<h2 id="引言一个困扰了-ai-二十年的老问题">引言：一个困扰了 AI 二十年的老问题</h2>
<p>在 Transformer 出现之前，主流的序列模型叫 <strong>RNN（循环神经网络）</strong>。</p>]]></description></item></channel></rss>